Mobility futurist, automotive thought-leader, and former Toyota executive James Carter recently interviewed Bright Way Vision executives on how Bright Way Vision and its GatedVision technology are paving the way toward a future of safe 24/7 autonomy.
James asked thought-provoking questions in his talk with Co-Founder & CEO Ofer David, Ph.D., and Co-Founder & EVP of Product and Business Development Eyal Levy. We've laid it all out in the following article, with the inclusion of spectacular images from our night camera system.
Q: How did you start?
We began the company as a spin-off of Elbit Systems, a defense technology giant, in 2011. In the early years, we developed a CMOS sensor pixel and sold our first camera, and by 2015, the company began taking off. Today we're an international company active in the truck, AV, and ADAS markets, working with most of the leading automotive brands worldwide and their T1s.
Q: What is your future mobility vision?
Autonomous and semi-autonomous vehicles are here to stay, and they’re set to become the reality in the near future. The next jump is clearly autonomy, and it may take some time to get to full autonomy evolving from today’s L2 to L4 and L5. Driverless trucks will transform the trucking industry, especially as the US trucking industry is struggling with a constant shortage of truck drivers. Currently, a typical truck is only actively being driven 35-38% of the time, leading to many wasted potential operational hours.
Moving to autonomy is compelling and makes strong financial sense. The idea of “hub-to-hub” transport is easier to handle and involves transport that relies on driverless technology on remote roads and allows a driver to take over through hubs located at the entrance to more populated areas. Hub-to-hub will be the first and most important step for autonomy in general.
Q: Can we trust an autonomous truck to brake in time?
Autonomous trucks will not have a driver or rely on human decisions in emergency situations. Their onboard systems must know how to respond and act and make decisions in every scenario, including extreme situations. One of the most common situations is driving at night, often in adverse weather, with a truck loaded with cargo on a highway. In these situations, it is critical to be able to identify every object on the road, especially those far away from the vehicle, in order to brake on time and safely enable any emergency stop.
The market’s real need is 170m as a minimum braking distance to give the truck the necessary response time to react to sudden changes, VRUs, and obstacles on the road.
To enable safe braking distances and for the benefit of all road users, autonomous vehicles will use a Bright Way Vision camera, which is the only product on the market that can see clearly ahead up to 250m away, in dark conditions and adverse weather. Bright Way Vision's camera is ideal for forward and backward-looking needs for trucks, ADAS, and AVs.
There are several ways to detect vehicles on the road and other large objects, but most solutions fail with small objects in darkness or complex weather situations - particularly when these situations are combined (adverse weather at night). Bright Way Vision developed the only method of detecting smaller objects with low reflectivity (including gray on gray ), such as tires, fallen cargo, motorcycle helmets, small animals, and more. These objects can't be detected by LiDAR or radar and, when hit, may cause real damage.
Q: How are non-drivable objects detected?
Our technology’s unique capability of detecting non-drivable objects solves a problem that no other visual solutions can address, as they lack the capability to detect small objects or offer clear views in darkness or adverse weather.
We also offer long-range stereo night vision and other solutions designed for long ranges up to 250m, and can even detect vehicles' retro-reflectors regardless of weather and lighting conditions.
Q: How does background removal equal better contrast?
When the environment is busy and focus is placed widely on multiple objects, sensors will often collect a lot of irrelevant and background data. This confuses the perception engine and hinders real-time decision-making. Our camera removes the background and improves the image contrast, giving a clearer image for visible and safe driving.
Once Bright Way Vision’s camera removes the background, pedestrians are detected as they stand out and the camera’s range of vision is left completely clear, leaving no chance of false positive alerts.
Q: What is the power of shadow detection?
Our system operates by gathering multiple layers of information, including bright images in low-lighting scenes. When illuminated from two different angles, the detection of even small objects with very low reflectivity is possible. This is done by analyzing the shadows cast by the obstacle.
Normally, extremely high resolution is required to identify such objects, but as we rely on shadows, and the shadows are way larger than the original object, we can detect these objects even with lower resolution. It may seem that with different points of illumination (left and right), the system will detect two shadows, but by comparing the images, we can observe which objects obstruct the road that wouldn’t be visible if the objects were only viewed from one side. To this very day, we haven't experienced a single false positive.
Current ADAS solutions become less usable in darkness or adverse weather conditions such as rain or snow. Rain is especially debilitating for most vision systems, and pedestrian fatalities in darkness are already a challenge, even to vehicles with human drivers.
Q: So what’s next for Bright Way Vision?
In the future, as we add new and exciting capabilities as well as full color to our camera system, we will begin to equip it to provide a highly effective daytime solution. So, our final goal is to provide an all-in-one vision system that can work as a replacement for camera systems entirely.
Q: What has been the response from OEMs?
We’ve received extremely positive feedback, time and time again, from the leading brands in the automotive industry, our partners in the journey. The one thing all our customers have in common and that we hear across the board from all OEMs is surprise and awe witnessing a drive with a clear image in the rain, fog, and darkness, in multiple use cases that have never been possible before.
But reactions have also been very diverse, as each OEM has its niche and way of operating. For example, some would like to share an image with the driver, while others don’t. Some OEMs are looking for a camera that can detect small objects at long ranges, some are interested in a standard-performing daytime camera, others are interested in the depth map, and some want to find a solution that performs well in low lighting and adverse weather conditions. When it comes to OEMs, what makes a product stand out depends on what the manufacturer is looking to achieve and which use-cases they want to resolve, safely.
Q: How does GatedVision work?
The GatedVision camera releases a certain amount of illumination for each range, which increases along with the range.
With most camera solutions, the image becomes less clear as the distance increases due to the degradation of returning light. We avoid this by using regular reflective objects to spread and return the light we put out. We output more light for farther distances, light that only our cameras can see, so it doesn’t obstruct or blind oncoming vehicles.
The reflectors act as an extra layer of information on top of our own and as data that adds thoroughness at the cost of very little computational energy, as it utilizes information that's already there. One of the common requests from trucking companies was a solution that makes it easier to merge into traffic. This requires seeing and clearly identifying the obstacles all around the vehicle at least at 150m distance, including behind the vehicle.
In the below example, you can see one light, which may look like a motorcycle in the HDR camera view, but our camera system clearly identifies it as a vehicle. In situations where a vehicle has both lights off, our system will also clearly identify it.
Q: How does GatedVision compare to thermal and LiDAR solutions?
The requirement for an automotive night vision camera is unique as you need the solution to address low lighting but simultaneously contend with high-energy beams from the headlights of other vehicles. To allow the sensor to work optimally, we use our own illumination, which allows us to control the range and prevent the computation unit from being overwhelmed with additional unnecessary information.
Another existing solution currently used to address low-lighting is thermal imaging. Still, in colder climates such as Canada and Northern Europe, it becomes useless as the temperatures are very low, and the differences between outside temperatures and objects on the road are minimal.
We provide the functionality of LiDAR, but while we don't have the depth resolution of this solution, neither does a human driver. Therefore, we focus on collecting the necessary information and leaving out unnecessary data.
VISDOM, a Night and All-Weather Camera
Our camera system, VISDOM, is powered by our original and patented GatedVision technology. The system is composed of a gated camera and illuminator, which is certified as a front position light. It can be incorporated into any of the front four light sources: headlight, fog light, front position light, or daytime running light. The key component is a small laser chip that emits illumination.
The camera is a separate small component that can be installed behind the rear-view mirror, dashboard, or even the roof of the vehicle, and the illuminator would be installed in the bumper area. The illuminator unit flashes approximately 200,000 times a second, making it too fast for the human eye to catch and preventing it from becoming disruptive. The sensor is in the camera and can also be integrated into the day camera system, making it easy for OEMs to integrate without making architectural changes.
We are proud to join 20 of the world’s top organizations and use our innovative tech to solve automated driving in variable traffic and weather at the AI-SEE project.
A recent study by the American Insurance Institute for Highway Safety found that AEB (automatic emergency braking) systems only prevent crashes and fatalities in the daytime.